The call always happens in the same week of the quarter. Week nine, maybe week ten. The founder has stopped doing outbound because they are stuck in customer calls and a fundraise. The two AEs are working inbound and the odd referral. Pipeline for next quarter looks thin in a way that everyone can see on the dashboard and nobody wants to name out loud.
So the CEO does the reasonable thing. They ask around, get three intros to outbound agencies, and by Friday they are looking at a proposal for a dedicated SDR pod at 7,000 dollars a month with a three month minimum. The deck promises 15 qualified meetings a month. The math works on paper. If even a third of those meetings turn into opportunities and one closes, the thing pays for itself.
Six months later the contract is dead, the AEs have stopped accepting agency meetings, and nobody wants to talk about it.
I have watched this play out at four companies now. It is not because agencies are frauds. Most of the ones I have worked alongside are staffed by people who work harder than the average in-house SDR. It fails because of what the company handed over, and what it kept.
The number that should scare you before you sign anything
SaaStr ran a poll of more than 1,200 people on this exact question. Only 7 percent said outsourced SDRs "really" worked for them. Another 26 percent said it "sort of" worked. That leaves roughly two thirds who got little or nothing out of the money they spent. Jason Lemkin's write-up of that result is worth reading in full.
Share of B2B companies who said an outsourced SDR program really worked, in a SaaStr poll of 1,200+ respondents. Another 26 percent said it sort of worked.
I want to be careful with that stat, because it is the kind of number people wave around to justify a decision they already made. A 7 percent success rate does not mean outsourcing is a scam. It means the failure mode is systematic, and if it is systematic then it is predictable, and if it is predictable then you can decide in advance whether you are in the 7 percent or the 67.
That is what this post is about. Not which agency to pick. Whether you should be picking one at all this quarter.
The real cost of an in-house SDR, which nobody quotes either
Before you can compare, you need the honest in-house number. Most founders I talk to have a figure in their head that is the base salary. That figure is wrong by roughly half.
A fully loaded US SDR runs somewhere between 125,000 and 150,000 dollars a year once you count everything. One 2026 breakdown puts the average at about 134,000 dollars: 75,000 base, 25,000 in variable at full attainment, 21,000 in employer taxes and benefits, and 8,500 in tools. That last line item surprises people. A single SDR seat across a CRM, a sequencer, a dialler, a data provider, and an enrichment tool adds up fast.
Then there is the part that does not show up in any spreadsheet: time.
The Bridge Group's 2025 sales development research, which covers 350-plus B2B companies, puts ramp to full quota at 5.5 months, with the first qualified meeting landing around month 3.2. Median SDR tenure is 1.9 years. Do that arithmetic and it is bleak. You spend roughly a quarter of a rep's entire tenure with the company getting them to full productivity, and annual SDR turnover in the US sits around 39 percent, with each departure costing somewhere between 40,000 and 90,000 dollars in replacement and lost output.
Now the agency side. Nearshore SDR pods run roughly 2,400 to 5,000 dollars a month per seat. US-based onshore dedicated reps through an agency run 7,000 to 10,000 a month and up. Pay-per-meeting deals sit around 150 to 600 dollars a booked meeting for mainstream B2B, and north of 900 for enterprise targets. Ramp is two to four weeks instead of five months, because the agency already has trained bodies and a working stack.
On year one cost alone, outsourcing wins by a wide margin. That is the entire pitch, and the pitch is not lying.
Cost is the wrong axis to make this decision on.
Outsourcing is cheaper in year one and everyone knows it. The programs that fail did not fail on price. They failed because nobody had proven the message before handing it to someone else to repeat 3,000 times a month.
The one question that actually decides this
Here it is. Has anybody at your company, founder or AE or you, personally booked ten meetings from cold outbound in the last 90 days using a message you could write down?
Not "we tried outbound." Not "we have a sequence in HubSpot." Ten meetings, cold, from a message that exists on paper and that someone else could read and run.
If the answer is yes, an agency can work. You are buying execution capacity for something you already know how to do. That is a fair trade and it is what outsourcing is genuinely good at.
If the answer is no, you are asking an agency to discover your positioning for you. They will not. They cannot. They do not sit in your customer calls, they have not lost the deals you have lost, and they do not know why the last three customers actually bought. They will do what any rational vendor does when handed a vague brief: run a broad, safe, high-volume play against whatever list they can build, and book meetings that technically match the criteria you gave them.
This is the mechanism behind the 7 percent. It is not agency incompetence. It is that most companies outsource the search for product-market fit in a channel, then measure the agency on meeting count, which is the only thing they can control.
What you are actually buying, and what you are not
An outbound agency sells four things. Only two of them are usually worth the money.
Capacity. Bodies who will send the emails and make the calls. Real value, especially when your alternative is a founder doing it between board meetings at 11pm.
Speed. Two to four weeks to live versus five and a half months to a productive in-house hire. Real value, and the single strongest argument for going this route when the quarter is already in trouble.
Infrastructure. Domains, inboxes, warm-up, data, dialler, sequencer. Some value, though less than it used to be. Setting up a technically clean sending infrastructure is a solved problem now, and if you are going to run outbound for more than a year you will want to own it anyway. I wrote about the details of that in the piece on cold email deliverability.
Strategy. This is the one on every agency's website and it is the one you should assume you are not getting. Not because they are lying, but because good outbound strategy comes out of pattern recognition across your won and lost deals, and they have no access to that. You have it. You are the only one who has it.
Here is the split I recommend to every client who asks. You own the ICP, the message, and the qualification bar. They own the sending, the calling, the list building against your criteria, and the follow-up. If an agency wants to own the first three, that is a red flag, not a selling point. If they refuse to work inside your CRM, walk away entirely.
The AI SDR question, since it is on every board deck now
You cannot write about this in 2026 without addressing it, so: AI SDR tools price at roughly 500 to 10,000 dollars a month depending on tier, and the honest all-in figure once you add data, infrastructure, and someone's time to run the thing lands closer to 900 to 12,000.
My read after building several of these systems: pure AI outbound erodes meeting quality in a way that shows up around week six. The volume looks great. The reply rate looks fine. Then you read the replies and half are polite confusion. What works is the hybrid, where the machine does research, list building, enrichment, and first-touch drafting, and a human does the qualification and the actual conversation. That combination is where the cost-per-opportunity improvements are real. I went deeper on where these tools hold up and where they fall over in the AI SDR reality check.
So the choice is not two-way. It is four-way, and most companies under 100 people are best served by some blend.
If you do outsource, run it like this
Assume you have answered the ten meetings question with a yes. Here is the operating model I would put in place before the contract starts, not after.
Write the qualification bar down before you sign
One page. What counts as a qualified meeting, in criteria an outsider can apply without asking you. Company size range, geography, the specific job titles, the trigger or situation that makes the timing right, and the disqualifiers. Especially the disqualifiers. "Not currently in a procurement freeze" and "not a competitor's customer inside contract term" save more wasted calls than any positive criterion.
If you cannot write this page, you are not ready. That is a genuinely useful test, and it costs nothing to run. Most of the work overlaps with defining your ideal customer profile properly, so if you have done that already you are most of the way there.
Pay for accepted meetings, not booked meetings
This is the single biggest change you can make and agencies will push back on it. When a vendor is paid on meetings booked, they have no financial reason to decline a borderline meeting, and borderline meetings get booked. Quality of leads is consistently the top complaint clients report about outsourced programs, and this incentive structure is why.
Restructure it. A meeting counts when the AE marks it accepted against the written bar, within 48 hours, with a reason code if rejected. Build in a dispute process so the agency is not at the mercy of an AE having a bad week. Then hold a weekly call where you review every rejected meeting together. That call is where the actual learning happens.
Make them work inside your CRM
Non-negotiable. Every account, contact, activity, and sequence step lands in your HubSpot or Salesforce instance, not in the agency's tooling. Two reasons. First, when the contract ends you keep the data and the learning instead of starting from zero. Second, you can measure them properly, which means source-to-close reporting rather than a monthly PDF of meetings booked.
I have seen companies run 14 months with an agency and end up with a spreadsheet. That is not a program, that is a rental.
Set kill criteria on day one
Write down, before you start, what result at day 90 means you stop. Something like: fewer than 8 accepted meetings a month by month three, or an accepted-meeting rate under 60 percent, and the contract ends. Put it in writing and share it with the agency. Good agencies appreciate this, because it tells them exactly what they are being judged on. The ones that get uncomfortable are telling you something.
When in-house is clearly the right answer
Three situations, and I am fairly firm on all three.
Your sales cycle is long and technical. If a first call requires understanding a regulated workflow or a technical architecture, an outsourced rep working three accounts alongside yours will not get there. The ramp problem that makes in-house expensive is exactly the thing that makes it necessary here.
Outbound is your primary channel for the next three years. At that horizon the year-one cost advantage of outsourcing inverts. In-house tenure improves productivity in years two and three, and you keep the institutional knowledge instead of renting it.
You have someone who can actually coach. This is the one people skip. An SDR without a manager who reviews calls weekly is a very expensive email-sending machine. If nobody on your team has run sales development before, hiring two SDRs is not building a team, it is buying two people the chance to fail slowly. The ramp and onboarding structure matters more than the hire itself, which is a whole topic in sales onboarding and ramp design.
And the honest fourth case, which is not about strategy at all: sometimes an agency is the right call simply because you need pipeline in this quarter and a hire will not produce a meeting until the next one. That is a real constraint and it is fine to act on it. Just do it with your eyes open, with the kill criteria written down, and without telling yourself it is a permanent solution.
How I would sequence it at a 50-person company
If someone handed me this problem tomorrow at a Series A with two AEs and no SDRs, here is the order.
Month one, the founder or the strongest AE runs 40 accounts by hand. Not a sequence blast, actual research and real messages. Target ten meetings. This is the test that decides everything downstream and it costs you nothing but time. If you cannot get ten meetings from 40 well-chosen accounts, the problem is the message, and no amount of outsourced volume fixes a message problem. It makes it worse, faster.
Month two, write down what worked. The exact trigger you looked for, the exact opening line, the objection that kept coming up. This document is the asset. It is also the thing you hand to whoever runs outbound next, in-house or not.
Month three, add capacity. Now the buy-versus-build question is a straightforward one about cost, speed, and how long you plan to run the channel, and you can answer it in an afternoon. Pair it with an AI research layer for list building and enrichment so whoever is sending is spending their time on conversations rather than tabs. That build is most of what we do in go-to-market engineering and AI automation work.
The failure I see is companies doing month three first, then never going back to do month one. And then concluding that outbound does not work for their market, which is almost never true. What did not work was buying execution before there was anything to execute. If you want the broader version of that argument, it is the same one I make about pipeline generation generally.
Not sure whether you are ready to outsource?
Book a free 30-minute call. We will look at your last 90 days of outbound and tell you honestly whether the problem is capacity or message.
Book a call →Frequently asked questions
How much does an outsourced SDR cost per month?
Nearshore pods typically run 2,400 to 5,000 dollars per seat per month. US-based onshore dedicated reps through an agency run 7,000 to 10,000 and up. Full-service programs including data, tools, and reporting land in the 4,000 to 12,000 range. Pay-per-meeting pricing sits around 150 to 600 dollars per booked meeting for mainstream B2B targets and above 900 for enterprise. Compare that to a fully loaded in-house SDR at roughly 134,000 a year, or about 11,000 a month, and outsourcing wins on year-one cost in almost every scenario.
Is an outsourced SDR cheaper than hiring in-house?
In year one, yes, usually by 40 to 60 percent once you count recruiting, benefits, tools, and the five and a half months of ramp before an in-house hire hits full quota. The gap narrows in years two and three as tenure improves in-house productivity. If outbound is a permanent channel for you, build. If you need pipeline this quarter or you are testing a new segment, rent.
Why do most outsourced SDR programs fail?
The most common cause is an ICP that was never properly defined or transferred, which pushes the agency toward broad, safe outreach. The second is paying on meetings booked rather than meetings accepted, which removes any reason to decline a borderline booking. The third is letting the agency work in their own tooling, so you end up with no data and no learning when the contract ends. All three are fixable before you sign and nearly impossible to fix after.
Should I use an AI SDR instead of an agency?
For research, list building, enrichment, and first-draft messaging, the AI layer is genuinely better value than paying humans to do the same work. For qualification and the actual conversation, it is not there yet, and pure-AI outbound tends to lose meeting quality after the first month or two. The hybrid setup, where AI handles targeting and drafting while one human owns the conversations, is where most companies under 100 people should land.
How long should an outsourced SDR pilot run?
Ninety days minimum, because two to four weeks of that is ramp and you need at least eight to ten weeks of real sending to read a result. But set the kill criteria on day one: a specific accepted-meeting count and accepted-meeting rate that determines whether you renew. Anything shorter than 90 days does not produce a readable signal, and anything longer without written criteria turns into a subscription nobody reviews.
Outsourcing SDRs is a capacity decision that companies keep making as a strategy decision. Get the message right first, in your own hands, on a small number of accounts. Then decide who runs it at volume. Do it in that order and you can be in the 7 percent. Do it in reverse and you will be writing the post-mortem in November.
If you want help figuring out which side of that line you are on, or you need the CRM and reporting layer in place before an agency starts, that is what we build. Get in touch.